Accuracy of analytic potentials for orbits of satellites around a Milky Way-like galaxy: comparison with $N$-body simulations

Abstract

To study the orbits of satellites, a galaxy could be modelled either by means of a static gravitational potential, or by live $N$-body particles. Analytic potentials allow for fast calculations, but are idealized and non-responsive. On the other hand, $N$-body simulations are more realistic, but demand higher computational cost. Our goal is to characterize the regimes in which analytic potentials provide a sufficient approximation, and those where $N$-bodies are necessary. We perform two sets of simulations using both Gala and Gadget, in order to closely compare the orbital evolution of satellites around a Milky Way-like galaxy. Focusing on the periods when the satellite has not yet been severely disrupted by tidal forces, we find that the orbits of satellites up to $10^{8} {\rm M_{\odot}}$ can be reliably computed with analytic potentials to within 5% error, if they are circular or moderately eccentric. If the satellite is as massive as $10^{9} {\rm M_{\odot}}$, errors of 9% are to be expected. However, if the orbital radius is smaller than 30 kpc, the results may not be relied upon with the same accuracy beyond 1--2 Gyr.